Quantification of Somatic Coffee Embryo
Growth Using Machine Vision

P.P.Ling, Z.Cheng, D.J.Musacchio

Abstract: Dynamic features were investigated to quantify somatic coffee embryo development
between the maturation and germination stages. An image registration algorithm was implemented
to register two images of the same embryos acquired at two different times. Embryo growth was
quantified using two features, elongation coefficient and growth aspect ratio, measured from 
registered images and related to the embryos' viability for germination. The approach was tested
using 462 embryos regenerated from various embryo genesis prototypes. In predicting embryo 
eventual germination, the machine vision system consistently outperformed the prediction made 
by an expert. The machine vision system's success rates ranged from 61.5 to 85.1% while the 
expert's success rates ranged from 43.1 to 69%.